altruios

24 ms·

421.
by altruios 3y ago
well... what we have an an LLM doesn't is state and feedback and restraint, a sense of time. We have an internal state of some kind, an LLM cheats having a state by having every input be the entire conversation + one new input. making
422.
by altruios 3y ago
So this is a chat-bot designed to help you self diagnose 'poor dating behavior'... my first question is... what's the source training data look like? My second question is how is this a good idea? My biggest worry about somet
423.
by altruios 3y ago
That hypothesis is exactly what they are testing though, no? What happens when you take away that experience. So the best way to think about it is to wait and see. I think I generally agree with your prediction, but I think it's becaus
424.
by altruios 3y ago
digital voting is the use-case I hear get brought up all the time... seems like it would work...
425.
by altruios 3y ago
data (the website) is being 'seen' by a model. training data is NOT the model. there is NO text from any website in any LLM model. The website is not being copied to an internal model anywhere, it is being looked (processed throug
426.
by altruios 3y ago
It would be perfectly legal that a million government agents went it coffee shops and recorded what they heard. It is the leaving of government property on private property that is the real issue, as well as transparency... not access of in
427.
by altruios 3y ago
Is it illegal to hallucinate and say words that happen to be verbatim a copy-write article you read once a blue moon ago? The key point is - this isn't being copied from the website, it's being generated from the model. this is no
428.
by altruios 3y ago
AI is merely exposing how insane copyright law is. Using public facing data for training is exactly as legal as looking at that data yourself. An AI regurgitating information is exactly as legal as a human doing it. If you are allowed to -
429.
by altruios 3y ago
Having no context does allow for a fresh perspective... Having to think things through slowly step by step may reveal errors other's glossed over... the cerebellum is important. But maybe, since we really know less about the brain than
430.
by altruios 3y ago
I am continually surprised by how much of the modern offices of America run on excel. My favorite feature is when you ever so slightly click to the edge of a cell in the middle of a 10,000,000 line csv file, it warps you to the bottom of th
431.
by altruios 3y ago
I figured the raw approach of diffusion-esque was tried (without techniques lifted from newer LLMs {chatGPT is only a year old}). I am thinking that those attempts were made largely before LLM's were popular. Self attention and positio
432.
by altruios 3y ago
LLM's generate 'the next token' require a serialization of data, and prevents 'too much' parallelization. I wonder if some sort of diffusion hybrid is being worked on. Something that approximates a complete text ans
433.
by altruios 3y ago
Glad that I never installed that. I have heard about these claims since tic-tok's inception. Glad there is actual data to back it up. This says nothing about data capture through the app: it is solely about artificially promoting or s
434.
by altruios 3y ago
This is wonderful. A fantastic step towards indexing video... I have some ideas I'd like to share - something to improve this project: 1: Are you only scraping the video id and text content? (the meta-data and description text is still
435.
by altruios 3y ago
The tone and energy you are sending is kinda sus. You are arguing you would rather treat them worse. Cruelty for the sake of it... because anything else would be 'undeserving burdens' on the 'honest working class folk'.
436.
by altruios 3y ago
The road (train of thought) is different from the pathfinding (emotional weight of thoughts), and an error in pathfinding does not indicate anything (error/traumatic/important/wrong) in a train of thought. This anxiety does n
437.
by altruios 3y ago
I don't disagree it's a human problem... But at the same time... There's the idea of a giftschrank around 'dangerous / false' information... That is a tool to combat the spread of information deemed dangerous,
438.
by altruios 3y ago
why should all computing be deterministic? let me show you this "genius"/"wrong-thinking" person as to say about AL(artificial life) and deterministic computing. https://www.cs.unm.edu/~ackley/
439.
by altruios 3y ago
I have rarely seen such heartless victim blaming on hackernews...
440.
by altruios 3y ago
maybe we just need a better system to measure 'trusted speech'? Censorship is a tool, but when all you have is one tool... well... We need better tools, not more censorship!
441.
by altruios 3y ago
My biggest feedback is in the UI information displayed. Before I try, before I 'log in', I have no way on the landing page - or navigating around a bit - to find out how much it costs. That knowledge up front would make me as a bu
442.
by altruios 3y ago
good metaphor. but extending it: do we have the equivalent tests for consciousness (I don't think so, right?)? So by all tests (1 - visual inspection), it passes as gold... We need better tests: to test consciousness... and better defi
443.
by altruios 3y ago
one small detail to clarify. We can't build a general 'halting determiner' for any arbitrary program. But to say that for any particular program a halting machine can be developed is a different proposition. A simple example
444.
by altruios 3y ago
Okay, you have me intrigued. Which studies? For rats or cats? Is this a 'cheat' to learn languages better, lol?
445.
by altruios 3y ago
Until we need to figure out how to reintroduce that 'critical period' property at our will.
446.
by altruios 3y ago
So what if we have an AI prompter-assister... we take an input prompt: run it through an AI to transform that prompt into proper AI talk - then that is what's prompted to the actual AI... That way the AI will always have the best promp
447.
by altruios 3y ago
36 years is a long time to make these headaches go away...
448.
by altruios 3y ago
I too was blind sided with the fact FTX was going down... Never even heard of them before they crashed. I would like to see a public accounting too!
449.
by altruios 3y ago
net harm in other parts of the world? Illegality is not directly corelated with harm, sadly. I mostly agree - but would argue its a lot more gray than black & white.
450.
by altruios 3y ago
whoa... hold on there... I replace my 2080ti with a 4080ti... is that's not an upgrade: that's instead categorically always a 'patch' to you? metaphysically - a replacement says nothing to the quality of replacement. Sil

More ›